radeon/llvm: Use correct float->int conversion opcode on SI.
authorTom Stellard <thomas.stellard@amd.com>
Wed, 5 Sep 2012 15:28:31 +0000 (11:28 -0400)
committerTom Stellard <thomas.stellard@amd.com>
Wed, 5 Sep 2012 17:17:17 +0000 (13:17 -0400)
commit12d3d6f6ab34d0ede8a2798fe68e7c9b1c894dac
tree8d76d4a111f86c2f52af224890f140673335af8a
parentd68e337c600a1b644d84e4703c92c2087aff82a4
radeon/llvm: Use correct float->int conversion opcode on SI.

V_CVT_I32_F32 converts floats to signed integers, but we were using
V_CVT_F32_I32 which convertes signed integers to float.
src/gallium/drivers/radeon/SIInstructions.td